THE CONNECTION BETWEEN DELTA SCUTI STARS AND CLOSE BINARY PARAMETERS A THESIS SUBMITTED TO THE GRADUATE SCHOOL

THE CONNECTION BETWEEN DELTA SCUTI STARS AND CLOSE BINARY PARAMETERS A THESIS SUBMITTED TO THE GRADUATE SCHOOL IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE MASTER OF SCIENCE IN PHYSICS BY GARRISON H. TURNER ADVISOR: DR. RONALD KAITCHUCK BALL STATE UNIVERSITY MUNCIE, IN JULY 011

Table of Contents List of Tables List of Figures iv v 1. Introduction 1 1.1 Background 1 1. Pulsating Stars in Close Eclipsing Systems 3 1.3 Effects of Binarity on Pulsation 4. Stellar Dynamics 6.1 The Dynamics of Close Binary Systems 6.1.1 Center of Mass and Potential 7.1. Roche Structure 9. Stellar Pulsation 11..1 Radial Pulsation 1.. Nonradial Pulsation 14..3 Properties of δ Scuti Stars 14 3. Observations 16 3.1 Target List 16 3. Summary of Observations 18 3..1 Telescopes and Equipment 18 3.. Observing Programs 19 3.3.1 Results 1 4. δ Scuti Stars in Close Binary Systems 4 ii

4.1 Literature Review 4 4. δ Scuti Stars With and Without Companions 46 4.3 Pulsation Period vs. Orbital Period 48 4.4 Potential vs. Pulsation 51 4.5 Roche Lobe Filling 53 4.6 Force and Pulsation 54 4.7 Resonance 57 4.8 Pulsation Period vs. Orbital Parameters without Uncertainties 59 4.9 Summary of Results 64 4.10 Theoretical Considerations 64 5. Conclusion 68 Appendix 71 References 75 iii

List of Tables Table 1 Targets for Observing Campaign 0 Table Results from Observing Campaign Table 3 General Properties of Close Binary Systems with Pulsating Component 44 Table 4 Pulsating Systems with Uncertainties 45 Table 5 P pulse /P orb ratios 57 Table 6 Summary of Fit Relationships 64 iv

List of Figures Figure 1 Pulsation Period vs. Orbital Period given by Soydugan et al. 5 Figure Binary System in Corotating Frame of Reference 7 Figure 3 Corotating Frame of Reference with Test Mass m 9 Figure 4 Light Curve for CZ Aqr on the night of 31-July-010 (UT) 3 Figure 5 Power-Spectrum for CZ Aqr with Peak at 0.0331 day 4 Figure 6 Power-Spectrum of CZ Aqr with 0.0331 Day Peak Removed 4 Figure 7 Data for CZ Aqr Phased onto the Single 0.0331 Day Period 5 Figure 8 Light Curve for EY Ori on night of 17-Nov-010 (UT) 6 Figure 9 Power-Spectrum of EY Ori with Peak at 0.1030 Day 6 Figure 10 Power-Spectrum of EY Ori after Removal of 0.1030 Day Period 7 Figure 11 Data for EY Ori Folded onto the 0.1030 Day Period 7 Figure 1 Data for EY Ori Folded onto the 0.061 Day Period 8 Figure 13 Light Curve for FO Ori on 3-Jan-010 (UT) 30 Figure 14 Power-Spectrum of FO Ori in the B Filter with Peak at 0.09 Day 30 Figure 15 Power-Spectrum of FO Ori in the B Filter with 0.097 Day Peak 31 Figure 16 FO Ori B Filter Data Phased onto the 0.09 Day Period 31 Figure 17 FO Ori B Filter Data Phased onto the 0.097 Day Period 3 Figure 18 Light Curve for QY Aql on the Night of 1-July-010 (UT) 33 Figure 19 Power-Spectrum of QY Aql data in V-filter with Peak at 0.1119 Day 33 Figure 0 Power-Spectrum of QY Aql with Peak at 0.1196 Day 34 Figure 1 QY Aql V Filter data Phased onto 0.1119 Day Period 34 Figure SY Cen on the Night of 31-May-010 (UT) in the V Filter 36 Figure 3 Power-Spectrum of SY Cen in V Filter 36 v

Figure 4 Light Curve for WY Leo on night of 0-March-010 (UT) in the V Filter 38 Figure 5 Power-Spectrum for WY Leo in the V Filter showing 0.0457 Day Period 39 Figure 6 Light Curve of X Tri in the V Filter on the Night of 6-Oct-010 (UT) 40 Figure 7 Power-Spectrum of X Tri in the V Filter Showing Peak at 0.00 Day 41 Figure 8 Data for X Tri Folded onto Single Phase for 0.00 Day Period 4 Figure 9 Histogram of δ Scuti Stars in Close Binary Systems 47 Figure 30 Histogram of Single δ Scuti Stars 47 Figure 31 Pulsation Period vs. Orbital Period (With Uncertainties) 49 Figure 3 Pulsation Period vs. Potential on Surface of Primary (With Uncertainties) 5 Figure 33 Pulsation Period vs. Percent of Roche Lobe Filled (With Uncertainties) 54 Figure 34 Pulsation Period vs. Force on Surface of Pulsating Component (W/Unc) 56 Figure 35 Pulsation Period vs. Orbital Period W/O Uncertainties 60 Figure 36 Pulsation Period vs. Potential on Surface W/O Uncertainties 61 Figure 37 Pulsation Period vs. Percent of Roche Lobe Filled W/O Uncertainties 6 Figure 38 Pulsation Period vs. Force on Surface of Pulsating Component W/O Unc. 63 vi

Introduction 1.1 Background Pulsating stars have intrigued astronomers since the late 1500 s. The first star observed to brighten and dim periodically was o Ceti. First observed by David Fabricius in 1595, the second-magnitude star eventually faded from view only to brighten back to its original magnitude. By 1660, it was established that o Ceti dimmed and brightened according to an 11 month cycle. This periodic cycle of dimming and brightening earned o Ceti the nickname Mira (wonderful). Although the first attempts to explain the changes were erroneous, the dawn of variable star astronomy had nonetheless arrived. Thousands of stars are now known to be variable, of which pulsation is only one mechanism. The most famous class of these stars being the Cepheid variables. Named for the first variable of this class found in 1784 by the young astronomer John Goodricke, δ Cephei, these stars change brightness on the order of days rather than months. Their brightness typically changes on the order of a magnitude or more, thus they are often among the most detectable, even in galaxies beyond our Milky Way. Other classes of pulsating variables have been discovered and categorized over the

years. The most important parameters for classifying these objects are period, change in magnitude and position on the H-R diagram. The δ Scuti class was one of the later types of pulsating variables to be found. Most of these stars display low-amplitude changes in brightness and short periods, on the order of hours although a subclass does exist where amplitudes can reach 0.30 magnitude or larger. These stars are generally still on or close to the main-sequence, crossing the instability strip. Hence, their spectral range is between A through F types. The prototype of this class is Delta Scuti which was discovered to be variable near the turn of the 19 th century. Since its discovery, hundreds of stars have been included in this category. Rodriguez et al. (000) published a comprehensive list of 636 variables, along with periods and amplitudes. This class is also one of the fastest growing classes of variables. New CCD technology is allowing photometric detection of low-amplitude variability, and Breger estimates that between 1/ and 1/3 of the stars at the main sequence/instability strip intersection show variability between 0.003 and 0.010 magnitude (Breger 000). With the advent of photometric campaigns such as Kepler, this family should only continue to grow substantially over the next decades. Another major class of variable stars is the eclipsing binary stars. An appreciable percentage of the stars in our sky are among binary stars. If the stars orbital plane lies in the line of sight with respect to observers from Earth, the stars will pass in front of each other twice in one orbital cycle. This is the famous eclipsing phenomenon. One such eclipsing star, Algol, was first observed to change brightness in 1670. An

explanation for the variation was to wait for over a century. Again, the young John Goodricke observed the magnitude changes and in 178 hypothesized a second body passing in front of the brighter component. There exists a subclass of binaries of which the individual components are not resolvable. These belong to a group called close binaries. Many of the components of these systems are separated by small orders of the sum of the stellar radii, generally 0 or less (Hilditch 001, pg 1). These short distances also lead to short periods, on the order of tens of days or less. All of the binaries in this thesis fall within these ranges of separation and period. Binary stars are important and eclipsing binaries even more so. Using spectroscopic techniques, the ratio of the velocities and thus the mass ratio can be obtained. Because the eclipsing phenomenon can be observed photometrically and the angle of inclination is known to a much greater certainty, other parameters such as individual masses, individual radii, and the ratio of their effective temperatures can be ascertained when both spectroscopic and photometric data are combined. Other orbital and component parameters able to be determined are the eccentricity of the orbit and an approximation of the percentage of the Roche Lobe filled for each component. 1. Pulsating Stars in Close Eclipsing Binary Systems The investigation of pulsating variables in close eclipsing binary systems is a relatively new area of study. The first δ Scuti-type pulsating variables in eclipsing 3

systems were found during the 1970 s with papers published throughout the decade. AB Cas was discovered to have pulsation in 1971 (Tempesti 1971), followed by Y Cam in 1974 (Broglia & Marin 1974). RS Cha and AI Hya were the last two of that decade (McInally & Austin 1977; Jøergensen & Grønbech 1978). The list grew slowly, reaching 5 by 006 (Soydugan et al. 006b). Most of these systems are classified as Algol-type eclipsing systems. This led Mkrtichian to introduce the term oscillating Eclipsing Algols (the term oea used hereafter; Mkrtichian 004). The number of systems used in this thesis is close to 40. Objects were taken from a paper in which 0 systems were collected and analyzed (Soydugan 006a) as well as publications reporting δ Scuti-type behavior in eclipsing Algol systems since 006. 1.3 Effects of Binarity on Pulsation A possible relationship between the orbital and pulsational periods was found by Soydugan et al. (006a). For their study orbital periods and pulsation periods were gathered from the literature for each system, along with other relevant orbital parameters such as mass and stellar radius. Using this information they generated a plot of the pulsation period vs. orbital period. This plot is replicated in Figure 1. They also investigated if a correlation exists between the force of the non-pulsating component on the pulsator and the pulsation period. However, only eight systems had enough relevant data published in the literature to perform the analysis. Even with a small sample, the authors claimed sufficient evidence of a correlation. In the course of doing undergraduate research on exoplanets, Dr. Ronald Kaitchuck and I discovered a δ Scuti star near DF Peg while performing differential 4

P pulse (day) 0.0 0.18 0.16 0.14 0.1 0.10 0.08 0.06 0.04 0.0 0.00 0 1 3 4 5 6 7 8 9 P orb (day) Figure 1. Pulsation period in days as a function of orbital period in days for 0 close binary systems. photometry. That discovery prompted an interest in pulsating stars, during the study of which a paper exploring orbital parameters and pulsation behavior on 4 systems was found (Tsvetkov & Petrova, 1993). In the paper the idea of a resonance between the orbital and pulsation period was mentioned. Investigating this concept further led to both the Soydugan et al. papers (006a; 006b; herein referred to as [1] and [] respectively). The first of these papers put forth the possible relationships between the orbital and pulsation periods and the force/pulsation correlation mentioned above. The second introduced a catalogue of close eclipsing binary systems in which at least one component fit the δ Scuti profile in terms of spectral type and mass. From these papers this master s thesis developed. It was the purpose of this project to observe targets from the catalogue and gather data from the literature in an attempt to further establish if binarity affects the pulsational behavior. 5

. Stellar Dynamics In this section some basic astrophysical principles are presented. Section.1 will focus on the dynamics of binary systems. Binary geometry, potential and Roche structure will be examined. In section. stellar pulsation will be considered along with the generalities of δ Scuti stars. While most of the information contained in this section is well-known, it is given here for the context of content in chapters 3 and 4, where observations of pulsating stars are presented and relationships between binarity and pulsation are obtained, respectively..1 The Dynamics of Close Binary Systems The first task is to formally define what is meant by a close binary system. R.W. Hilditch asserts we should interchange the term close with interacting to suggest that a close binary system is one in which the two stars are close enough to have their evolution affected by the presence of the companion (Hilditch 001; pg. ). Typically in the literature close implies the separation is on the order of tens of solar radii or less. For the purposes of this study, close fits both of these definitions. Of all the systems analyzed, the one with the longest orbital period is FO Ori with 18.8 days. Using Kepler s Third Law in the form

P 4 G( M M 1 a ) 3 (.1) where P is the orbital period, G is Newton s gravitational constant, M 1 and M are the masses of the respective components of the binary, and a is the semi-major axis. For this study M 1 and M will be denoted as M p and M s for primary and secondary (while the majority of pulsators included herein are considered the primary component; higher temperature and usually more massive, this is not exclusively the case). This leads to a separation for FO Ori of about 46 solar radii. This result is within tens of solar radii and hence within the assumptions of much of the literature. To describe any system in a classical manner, three variables are needed: distance, mass, and time. It is typical in the literature of binary systems to report masses and distance in units of solar masses and solar radii, respectively. In the area of pulsating variables, periods are often reported in terms of days or in frequencies of c/d, or cycles per day. These are the units of mass, distance and time that will be utilized throughout the rest of this paper. To this end, in 3 R Sun all calculations, the value for G used is given as 943.7 M day Sun (see appendix for the calculation of the conversion)..1.1 Center of Mass and Potential Y CM X M 1 M r 1 r Figure. A Binary system with components M 1 and M. The Center of Mass (CM) is at the origin. r 1 and r denote the distance from each component to the CM. 7

Figure.1 shows two stars, designated as M 1 and M situated distances r 1 and r respectively from the center of mass (the relative sizes of M 1 and M are to indicate mass; it is common for the more massive star to have the smaller radius). The center of mass is shown at the origin of the x-y plane. Using the magnitudes of r 1 and r with the equation of an ellipse it can be shown that binary stars have the following property: M 1 r (.). M r 1 The r 1 and r vectors are defined as the distance from the center of the star to the center of mass, thus the sum of the vectors provide the distance between the centers of both stars, or the semi-major axis of the system. Having defined the center of mass and the distance vectors, it is possible to now describe the gravitational potential. Figure. shows a test mass m a distance r away from the center of mass of the system. The gravitational potential energy of a particle of mass m a distance s from a body of mass M is given by the well-known equation Mm U G (.3). s If a second body of mass M is added, the potential energy can be found by superposition of the two bodies. Because the coordinate system in Figure.1 is a corotating frame of reference (objects fixed with respect to the center of mass), we also must include a centrifugal term: U c 1 m r (.4). Marking each mass with subscripts to differentiate between them and adding the centrifugal term we arrive at the total potential of the system: 8

m Y s s 1 M 1 r M X CM Figure 3. A corotating frame of reference showing a test mass m. U M 1 G( ) m r (.5). s 1 m M s m 1 If, finally, this result is divided through by the test mass m, the effective gravitational potential, Φ, is obtained. The result is M 1 G( ) r (.6). s 1 M s 1 This result of the effective gravitational potential will also be used in Chapter 4 to investigate if the potential in a binary system affects the period of pulsation. When the potential is analyzed, some important points in the x-y plane can be found. If we let the test mass rest on the x-axis and differentiate the potential, we find specific points where the derivative (the force) goes to zero. These are the Lagrangian points. Of particular importance for binary stars is the L 1 point. This point lies directly between the two components and is the point at which mass will flow if and when mass transfer occurs. 1.1. Roche Structure If the effective gravitational potential is calculated at a point in the x-y plane, the value 1 For a detailed discussion on the derivation of the effective gravitational potential, see chapter 18 of Carroll & Ostlie (007). 9

belongs to a group of points of which all share that same value. The surfaces these points make are called equipotential surfaces. Likewise, in three dimensions there exists a surface that encloses a volume and the points on the surface of this volume are also equipotentials. Close to each individual star, these surfaces are nearly spherical. Moving outward, they become less spherical and more tear-drop shaped. The L 1 Lagrangian point is where the tear-drop shapes for each star meet, also called the inner Lagrangian point. The largest tear-drop shape for each star is called the Roche limit. The Roche limit is the maximum volume a star can have and retain its own constituents under its own gravitational control. This maximum volume is also referred to as the Roche lobe. An important quantity involving the Roche lobe is called the effective radius. The effective radius is the radius a sphere would have if it occupied the same volume as the Roche lobe. Several have worked to develop approximations for this quantity; perhaps the most useful is that developed by Eggleton (1983) which is good to within 1% for all mass ratios. This approximation depends on the mass ratio q (defined as M 1 /M, or M /M 1 depending on which radius is being calculated). The formula is given as: R L a / 3 0.49q (.7) / 3 1/ 3 0.69q ln(1 q ) where q=m /M 1 is used to evaluate R L and reversed to evaluate R L1. One more useful approximation involves finding the equatorial radius of the Roche lobe in the y-direction as seen from the positive z-axis looking down on the x-y plane. This approximation was developed by Plavec & Kratochvil (1964) and is given as: R L ( eq) 0.084 (.8) a 0.378q 10

where the q in this instance is for M /M 1. With this q-value the equation gives R L (eq) for M (let q = M 1 /M to obtain R L (eq) of M 1 ). This result will play an integral role in Chapter 4 when investigating the potential on the surface of the pulsator. One important note should be made about the foregoing analysis; the orbits are assumed to be circular or at least at a very low eccentricity (which are assumed throughout the rest of the paper). While this assumption may not be explicitly justified, numerous estimates place the time for circularization in systems with orbital periods of less than 8 days at about 10 6 years, which is short enough that circularization should have taken place in most of the systems. Exceptions would obviously be the longer period systems, such as FO Ori and EY Ori. Results in Chapter 4, however, will give an argument for the circularization of at least FO Ori.. Stellar Pulsation The variability in the light curves of pulsating stars is due to periodic departure from equilibrium. Oscillations both deep inside the star and closer to the surface can cause departure from equilibrium that will, in turn, cause the luminosity changes detected in the light curves. Pulsations can be divided into two groups. Radial pulsations maintain the spherical symmetry of the star, while non-radial pulsations in general do not maintain the symmetry. Until recently, relatively few stars have been observed to pulsate. Carroll and Ostlie (007, pg. 496) estimate that only one out of 10 5 stars show pulsation. This ratio is likely to tend to unity as better detectors are produced and more precise photometric techniques are developed. New photometric campaigns, such as Kepler, should be able to provide a better estimate of the true fraction of stars in the sky that pulsate at amplitudes currently See Hilditch (001) for a detailed discussion on Roche structure and circularization in close binary systems. 11

photometrically detectable. One more reason why this ratio should increase is we now know what types of stars are likely to pulsate and which are not. Pulsations are driven by ionization zones. The main zones are the hydrogen partial ionization zone and the He II partial ionization zone. The He partial ionization zone is deeper in the star than the Hydrogen; however they both move toward the surface as T eff increases. A general description of the pulsation mechanism follows. Stellar layers are subject to compression from perturbation. The opacity of layers within a star are governed by Kramers law, which states that the opacity, κ, is given by T 3.5, where ρ and T are the density and temperature of the layer. Typically, the opacity decreases upon compression (Carroll and Ostlie, pg. 496; Cox 1980, pg. 137). However, for pulsations to occur, the opacity must increase so that some of the energy is damned up (otherwise energy leakage would not cause pulsational instability). In the ionization zones, the condition for an increase in opacity upon compression is found. These layers experience a phenomenon called temperature freezing (Huang & Yu 1998, pg. 486). During compression, the density increases and because the temperature is frozen the opacity thus increases. This allows energy in the layer to go into ionizing the atoms in these regions. Expansion in these regions occurs when enough energy has been built up. Once the layers expand they then release the heat and as the layer cools, the atoms recombine and the layer contracts to begin a new cycle. This is known as the κ- mechanism...1 Radial Pulsation Radial pulsation is characterized by sound waves traveling in the stellar interior. Simple models show radial pulsation periods are inversely proportional to the mean density of the star. 1

This is the well-known period-mean density relation. This model says that lower-density stars have longer pulsation periods, while stars with higher densities have shorter periods. Advanced models and observations also share this result. The fundamental mode of radial pulsation can be thought of as a sound wave with a node at the center and an anti-node at the surface. A simple linearized model for the fundamental period is given in Carroll and Ostlie (007, pg. 50) as: 4 G 0 (3 4) 3 (.9) where Π is the period, ρ 0 is the mean density, and γ is the ratio of specific heats of the gas. Although this model was built on simplified assumptions, using the density of known Cepheid variables produces periods consistent with observations. This model is also confirmed observationally in that it predicts stars with higher densities to have shorter periods, which is in fact what is observed. Higher mode pulsations are modeled with one or more nodes between the center of the star and the surface. As the waves travel through more nodes, the shorter the pulsation period becomes. Also, as more nodes are present, the amplitudes generally decrease. Typically, only low-order radial pulsations are detectable with current photometric techniques. In general, the majority of high amplitude and long-period variables pulsate radially; usually in either the fundamental mode or the first harmonic. Another way to write the period-mean density relationship is to invoke the pulsation constant, Q. In this form the relationship is given as Π ( / SUN ) 1/ =Q. This says that the 13

pulsation constant is equal to the period of the pulsating variable multiplied by the square root of the ratio of the mean densities of the variable and the sun, respectively... Nonradial Pulsation Whereas radial pulsations conserve radial symmetry, some stars pulsate such that certain regions of the stellar surface expand while others contract. As a result the spherical symmetry is broken. Such stars are said to exhibit nonradial pulsation. To describe nonradial pulsation, spherical harmonics are utilized. The numbers l and m are used to describe the surface of a sphere. l has non-negative values and m can be equal to any integer between the values of -l and +l. l represents the number of nodal circles, and there are m circles running through the poles, while l - m circles run parallel to the equator. If l = 0 then m = 0 as well, and the pulsation is purely radial. For a non-rotating star not gravitationally bound to a companion, description using spherical harmonics may prove to be difficult due to the lack of an identifiable pole. However, in the case of pulsating variables in circular orbits with a companion, the axis of rotation defines the pole. There are two main types of nonradial oscillations. The first are called p modes; p standing for pressure. Here pressure provides the restoring force for the waves traveling in the stellar interior. The second type are called g modes. Here internal gravity waves moving through the stellar interior provide the restoring force for the sound waves...3 Properties of δ Scuti Stars Most of the oea s encountered in the literature are of the δ Scuti family. These pulsating variables are found at the intersection of the instability strip and the main sequence on the H-R diagram. Because almost all of them inhabit this area, they have a specific spectral 14

range between about F8 and A (Templeton 010) although A5 is fairly hot (e.g. HD 0838 which is an A5; Turner & Kaitchuck 008). Most δ Scuti stars pulsate nonradially, although a few pulsate in radial modes alone (Breger 000). Because of their spectral range and position on the H-R diagram, they have higher densities than larger amplitude variables such as classic Cepheids. As a result their periods are significantly shorter. The typical range for periods is between 0.0 and 0.5 days (or about half an hour to about 6 hours) (Breger 000). Several overviews of basic properties of δ Scuti stars are available, i.e. Templeton (010), and Percy (007). A detailed description of the current state of δ Scuti stars is given by Breger (000). Because most δ Scuti stars exhibit nonradial pulsation, and oea are typically δ Scuti stars, we can regard the pulsations analyzed in Chapter 4 as generally nonradial in nature. It is rather curious that to date nothing in the literature indicates the discovery of larger amplitude variables such as Cepheids or RR Lyrae stars in these close binary systems. There may be a physical limitation to the development of these types of variables in close systems due to the proximity of the companion. However, if they existed, it is plausible that they would be the first discovered due to their large variability. This is not the case. The rate at which small amplitude variables (and by extension variables in places such as close binary systems) is increasing due to such programs as HIPPARCOS, MACHO, and OGLE (Breger 000). With the dawn of the Kepler age, the rate of these discoveries should only accelerate. 15

3. Observations 3.1 Target List All of the targets for this research project were taken from Soydugan et al. (006b). The targets were chosen based on factors such as magnitude, length of time available to observe the target, and position in the sky. For this study 15 targets were chosen with possible pulsation detected in six of those targets (FO Ori was observed for a different research project but periodicity was detected once the data were analyzed). This suggests photometrically detectable pulsation may be present in these systems on the order of 30-40 %. This is in contrast to two other campaigns working from the same catalogue. Two papers published in 009 reported successful detection of pulsation on the order of 10 % in these systems (Dvorak, 009; Liakos, 009). Each study observed over 0 systems. This 10% success rate may be due to observational limits. Therefore many of the targets in the catalogue already published in the literature without detection of pulsation should be looked at again by observers with different equipment than those of the author(s). X Tri was observed by Liakos et al. (009) and pulsational variability was not found in their study. However, in taking data for this thesis, it was discovered to have pulsation. As a result, observers should continue to observe systems for which no pulsation has been detected as pulsation may simply be below the detection

capabilities of the publishing author. Multiple observing campaigns are encouraged to confirm or deny detectable pulsations. Three other interesting cases should be noted. Observations of BG Peg had already been taken for purposes of this thesis before realizing it had already been published in the literature (Dvorak, 009). Because no significant difference between the published period and the period found in my data could be found, no further observations were taken and the results have been excluded from this analysis. WY Leo, on the other hand, while published in the same study as BG Peg, was found to have significantly different periods. As I only gathered one night of observations on WY Leo, compared to the 7 gathered by Dvorak, caution is encouraged before any claims of pulsation period changes in the system are made. However, it may be worthwhile to gather more data on this system to determine if a period change has occurred or is still changing. FO Ori was observed on 3 nights in January, 010, for an un-related study and subsequent analysis showed pulsational behavior. Therefore, although it is not in the catalogue, it was included in this study and as will be shown later, may be one of the more interesting data points for the purposes of this study. All of the data taken for this thesis was done using differential photometry. Because the typical star in the δ Scuti region on the HR diagram is of spectral type F0-A5, the best filters to use to detect pulsations are the B and V filters. Some studies restrict their observations to the B filter alone. This was not done for this study as the Ball State observatory is not optimized for observations in the B filter and many stars that might be used for comparison might be brighter in the V filter. However, the R and I filters were excluded (pulsation amplitudes are filterdependent with the B and V filters providing the highest amplitudes). Images were reduced 17

using the IRAF ccdred package, correcting for bias (underlying noise levels from the CCD), dark (thermal) and flat-field (pixel to pixel variations and impure optical) effects. Differential photometry was performed with the AIP4Win software package. The results of which were used to generate the light curves using Microsoft Excel. All light curves that indicated eclipse phenomena were fit with polynomials to flatten the data set before the period analysis was performed. It should be noted no attempt at pulsation mode identification has been made for the purposes of this thesis. Photometric errors were obtained in AIP4Win by performing a signal-to-noise calculation which included camera read noise, the gain of the camera, dark currents, and the sky background for the variable and comparison star(s). These errors were then added in quadrature to obtain the error for each individual differential measurement. 3. Summary of Observations The observations for this project were taken between January of 010 and December of 010. The following sections relate the observations in detail. The telescopic equipment and CCD cameras are given first. The list of targets with all relevant observational data is also presented. Finally the results for positive possible pulsation detection are given. 3..1 Telescopes and Equipment Data were collected with 4 different telescopes. The Ball State University Observatory houses two of the instruments. These include 0.4-m and 0.3-m diameter telescopes. Ball State is also part of the Southeastern Association for Research in Astronomy which operates a 0.9-m telescope at Kitt Peak National Observatory and a 0.6-m telescope at Cerro Tololo in Chile. The SARA telescopes are of the Cassegrain design, while the BSU scopes are Schmidt-Cassegrains. The 0.4-m telescope at Ball State University is a Mead LX 00 with an SBIG ST-10 CCD 18

camera attached. The telescope has a focal ratio of f/6 and a plate scale of 0.58 arc sec per pixel. The 0.3-m telescope is a Celestron with an SBIG STL-6303 CCD camera. This telescope is at an f/11 ratio, giving a plate scale of 0.47 arc sec/ pixel. The 0.9-m telescope at KPNO has an Apogee U4 CCD camera with an f/7.5 ratio, giving a plate scale of 0.4 arc sec/ pixel. The telescope at CTIO is equipped with an Apogee Alta CCD camera at a focal ratio of f/13.5 giving a plate scale of 0.6 arc sec/pixel. 3.. Observing Programs Table 1 describes the observing program for each star. Included are the comparison stars chosen, UT and Julian Dates for each observation session, which telescope was used, and the number of images taken in each filter. Coordinates for each were obtained from the SIMBAD database when possible. In all other cases coordinates were obtained from The SKY 6 (Bisque Software, Inc. 004). The Julian dates given are heliocentric-corrected dates for each night of observations. Once generated each light curve was processed with the PERANSO (Vanmunster 007) period analysis software package for periodic behavior on δ Scuti time scales using the Lomb- Scargle method (Lomb 1976; Scargle 198). Each data set was period-searched between 0.01 and 0.3 days, which encompasses the accepted range of periods found in δ Scuti stars. A general rule was followed that if a night of observations on an object did not indicate periodicity then it was labeled as a low priority target. Ideally an object should be looked at for a minimum of four hours before discounting variability (δ Scuti stars have a maximum period of a little less than 8 hours thus at least half of a full possible period should be gathered to fully discount the presence of pulsation). However this was not always done as variables in the 19

Target Comp Check UT Julian Instrument B Images V Images RA RA RA 455000 Target DEC DEC DEC AD Her 18 h 50 m 00.3 s 18 h 50 m 13.3 s 18 h 50 m 18.5 s 5-Jun-010 37.701 0.3-m ---- 7 +0 o 43' 16.5" +0 o 45' 01.6" +0 o 45' 30" 8-Jun-010 375.7534 0.9-m 86 86 AT Peg h 13 m 3.5s h 1 m 47.9s h 1 m 56.4 s 19-Aug-010 47.659 0.3-m 50 50 +08 o 5' 30.9" +08 o 33' 1.4" +08 o 31' 14.8" CZ Aqr 3 h m 0.6 s 3 h m 4.0 s 3 h m 34.6 s 31-Jul-010 408.763 0.6-m 84 84-15 o 56' 0.4" -15 o 59' 14.5" -15 o 55' 0.4" 8-Oct-010 477.7341 0.6-m 89 84 13-Oct-010 48.507 0.6-m 45 ---- EE Peg 1 h 40 m 01.9 s 1 h 40 m 17.3 s 1 h 39 m 04.3 s 18-Jun-010 365.806 0.3-m ---- 48 +09 o 11' 05.1" +09 o 00' 34.3" +09 o 03' 3.0" EG Cep 0 h 15 m 56.8 s 0 h 14 m 43.9 s 0 h 17 m 5.9 s 17-Aug-010 45.7196 0.3-m 31 31 +76 o 48' 35.7" +76 o 43' 11." +76 o 40' 07.4" EY Ori 05 h 31 m 18.4 s 05 h 31 m 0. s 05 h 30 m 59.6 s 17-Nov-010 517.7899 0.6-m 464 ---- -05 o 4' 13.5" -05 o 37' 7.5" -05 o 39' 14.6" 10-Dec-010 540.747 0.9-m 107 FO Ori 05 h 8 m 09 s 05 h 8 m 35.4 s 05 h 8 m 19.7 s -Jan-010 199.587 0.9-m 199 199 +03 o 37' 3" +03 o 38' 45.0" +03 o 36' 5.9" 7-Jan-010 04.606 0.9-m 168 04 10-Jan-010 07.6431 0.9-m 19 180 HS Her 18 h 50 m 49.7 s 18 h 50 m 1.7 s 18 h 51 m 07.7 s 18-Jun-010 365.664 0.3-m ---- 74 +4 o 43' 11.9" +4 o 38' 31.7" +4 o 38' 58.3" QY Aql 0 h 09 m 8.8 s 0 h 09 m 5 s 0 h 09 m 11 s 1-Jul-010 378.7433 0.6-m 60 60 +15 o 18' 44.7" +15 o 17' 34" +15 o 1' 49" 15-Jul-010 39.645 0.4-m ---- 70 9-Sep-010 448.6848 0.4-m ---- 96 RR Vul 0 h 54 m 47.6 s 0 h 54 m 37.7 s 0 h 54 m 48.8 s 6-Oct-010 475.676 0.4-m ---- 60 +7 o 55' 05.7" +7 o 53' 11." +7 o 57' 54." 7-Oct-010 476.564 0.4-m ---- 104 SW CMa 07 h 08 m 15. s 07 h 07 m 54.9 s 07 h 07 m 53.8 s 7-Apr-010 93.5004 0.6-m ---- 318 - o 6' 5.3" - o 3' 6.1" - o 5' 57." SY Cen 13 h 41 m 51.5 s 13 h 4 m 04.6 s 13 h 41 m 5.1" 31-May-010 348.4475 0.6-m ---- 16-61 o 46' 10.1" -61 o 44' 58.1" -61 o 44' 35." UX Her 17 h 54 m 07.8 s 17 h 54 m 06.6 s 17 h 54 m 13.0 s 5-Jun-010 37.6583 0.4-m ---- 93 +16 o 56' 37.8" +16 o 58' 14.8" +17 o 04' 37.8" V805 Aql 19 h 06 m 18. s 19 h 06 m 16.3 s 19 h 06 m 1.6 s 18-Jun-010 365.7374 0.4-m ---- 50-11 o 38' 57.3" -11 o 35' 39.9" -11 o 33' 18.3" WY Leo 09 h 31 m 01.1 s 09 h 30 m 5.5 s 09 h 30 m 54.1 s 9-Mar-010 84.6481 0.9-m ---- 115 +16 o 39' 5." +16 o 33' 8.0" +16 o 35' 46.1" X Tri 0 h 00 m 33.7 s 0 h 00 m 30.6 s 0 h 00 m 37.8 s 6-Oct-010 475.7550 0.4-m ---- 101 +7 o 53' 19." +7 o 47' 05.9" +7 o 55' 10.8" 7-Oct-010 476.6976 0.4-m ---- 1 Table 1. Given are coordinates for the targets and comparison stars, dates, instruments, and number of exposures. 0

observing runs (i.e. clouds or priority given to other targets that did show signs of pulsation) played a factor in which systems were focused on. The systems AD Her, AT Peg, and EG Ceph did not meet the 4 hour requirement. Nonetheless, for the data acquired on those systems PERANSO did not give positive results for periodic variability. SY Cen showed periodicity. However, this is based on one night of observations. Only one usable night of telescope time at Cerro Tololo was allocated to the project while SY Cen was observable. Thus, this object, above the others, is recommended for further observations to determine if periodicity does exist. On the other hand, RR Vul s light curve indicated possible periodicity from visual inspection. However, PERANSO did not detect a dominant period to sufficient confidence levels. Again, further study on this object is encouraged in order to confirm the absence or presence of pulsation. An object was determined to have strong periodicity if the period window in PERANSO showed a dominant peak above the noise in the Lomb-Scargle statistic (theta) and if this peak was not associated with a peak in the spectral window (the peaks in the spectral window show how the data were sampled). Only peaks in the period window that were not found to be artifacts of the observing schedule were deemed as a peak possibly due to stellar pulsation. The pre-whitening function was also used to remove the dominant peak in order to look for evidence of multi-periodicity. 3.3.1 Results The results of positive detection for periodic behavior are presented in this section. Of the observed systems, six were found to have strong periodicity and have not been published in the literature. WY Leo was found to have periodicity but different from that already published 1

Target POrb PPulse Amplitude (day) (day) (mmag) AD Her 9.7666 ------- ------- AT Peg 1.1461 ------- ------- CZ Aqr 0.868 0.0331() 1 EE Peg.68 ------- ------- EG Cep 0.5446 ------- ------- EY Ori 16.7878 0.1030(0) 0 FO Ori 18.8006 0.09() 5 HS Her 1.6374 ------- ------- QY Aql 7.96 0.1119(4) 1 RR Vul 5.0507 ------ ------- SW CMa 10.09 ------- ------- SY Cen 6.6314 0.0936(119) 1 UX Her 1.5489 ------- ------- V805 Aql.408 ------- ------- WY Leo 4.9859 0.0457(8)*????? X Tri 0.9715 0.0(1) 10 Table. Given are the orbital and pulsation periods, and the amplitudes. The asterisk by the pulsation period of WY Leo denotes that the pulsation period differs from that found by Dvorak (009), while the question marks in the amplitude column indicate the amplitude was not constant throughout the data set (see Fig. 4). (see above comments). Table gives the orbital period of each system observed, the dominant pulsation period (if detected), and the measured amplitude. All amplitudes were measured with the half-amplitude method. The numbers in parentheses represent the uncertainty on the measurements as determined by PERANSO. Each system will be discussed in detail with sample light curves and power-spectra given. In each case a description of the results of each system in which pulsation was detected is given with all associated figures relegated to the end of this chapter (error bars are included on all light curves).

CZ Aqr CZ Aqr was observed on three nights, 31-July, 8-Oct, and 13-Oct-010 (UT). A total of 18 images in the B-filter and 164 in the V-filter were acquired. The light curve in the B- filter for 31-July-010 (UT) is shown below in Figure 4. The photometric errors for the night ranged between 1.8 and millimag. The results of the period search are shown in Figure 5. The determined period is 0.0331 + 0.000 days. Figure 6 shows the power spectrum after the removal of the 0.0331 day peak. It is possible these residual peaks are indicative of CZ Aqr having multiple modes of pulsation. However, further observations will be needed to confirm whether or not CZ Aqr is indeed multi-modal at current photometric levels. Figure 7 shows the data for the 3 nights in the B-filter phased onto the 0.0331-day period. 1.03 1.0 1.01 Δ mag 1 0.99 0.98 0.97 55408.74 55408.77 55408.80 55408.83 55408.86 55408.89 55408.9 55408.95 JD 400000 Figure 4. Light curve for CZ Aqr on the night of 31-July-010 (UT). 3

Figure 5. Power-spectrum for CZ Aqr with peak at 0.0331 day. Figure 6. Power-spectrum of CZ Aqr with the 0.0331 day peak removed. Residual is 0.035 day. 4

Figure 7. Data for CZ Aqr phased onto the single 0.0331 day period. EY Ori Observations on EY Ori were performed on 17-Nov-010 and 10-Dec-010 (UT). A total of 571 images were acquired in the B-Filter over the course of the two nights. The light curve for the night of 17-Nov-010 is shown in Figure 8. The amplitude is on the order of 0 millimag while the photometric errors ranged between 3.0 and 4.0 millimag. The period was determined to be 0.1030 + 0.000 day. Figure 9 shows the power-spectrum with the 0.1030 day period, while Figure 10 shows the power-spectrum with the 0.1030-day period removed. The residual shows a peak at 0.061 + 0.0007 day and the data folded onto a single phase for both the dominant and secondary periods are shown in figures 11 and 1, respectively. The phase diagram for the secondary period is suggestive of multi-modal behavior. 5

-.55 -.54 -.53 -.5 Δ Mag -.51 -.5 -.49 -.48 55517.76 55517.8 55517.84 55517.88 55517.9 55517.96 55518 -.47 JD +400000 Figure 8. Light curve of EY Ori on night of 17-Nov-010 (UT). Figure 9. Power-spectrum for EY Ori with peak at 0.1030 day. 6

Figure 10. Power-spectrum of EY Ori after removal of 0.1030 day period. Residual is 0.061 day. Figure 11. Data for EY Ori folded onto the 0.1030 day period. 7

Figure 1. Data for EY Ori folded onto the 0.061 day period after removal of the 0.1030 day periodicity. FO Ori Data on FO Ori were acquired over the course of three days in January, 010 with the 0.9-m telescope. FO Ori was originally observed as part of a separate project at Ball State University by faculty member Dr. Ronald Kaitchuck and students, Joe Childers, Ken Moorehead, Danielle Dabler, Matthew Knote, Elizabeth Beal and me. However, after periodicity at low photometric levels was discovered, it was added to the study of this thesis. It was deemed a potentially important data point as the original Pulsation Period vs. Orbital Period relationship 8

given by Soydugan et al. did not have orbital periods beyond 10 days. FO Ori s 18.8 day orbital period thus serves as an interesting check case for the validity of any such relationship beyond periods of 10 days. The implications for the validity of a Period/Period relationship will be given in Chapter 4. Analysis of the 496 images in the B filter revealed a dominant period of 0.09 + 0.000 day and a secondary period of 0.097 + 0.000 day (the 583 images in the V filter give 0.089 + 0.000 day for the dominant period, thus in good agreement; although more images were taken in the V filter, the B filter data was slightly cleaner consistently). The light curve for the night of 3-January-010 in the B-filter is shown in figure 13. Figure 14 shows the powerspectra in the B filter for the three nights giving the dominant 0.09 day. Figure 15 shows the 0.097 day period with the 0.09 day period removed. Thus, again multi-periodicity is implied. Data folded onto single phases of each period can be seen in Figures 16 and 17, respectively. The amplitude of pulsation is about 50 millimag while the photometric errors were consistently (within ten percent) one millimag. 9

1.10 1.08 1.06 1.04 1.0 Δ Mag 1.00 0.98 0.96 0.94 0.9 0.90 55199.57 55199.67 55199.77 55199.87 JD +400000 Figure 13. Light Curve for FO Ori on 3-Jan-010 (UT). Figure 14. Power-spectrum of FO Ori in the B filter showing peak at 0.09 days. 30

Figure 15. Power-spectrum of FO Ori in the B filter with 0.097 day peak after 0.09 day peak removed Figure 16. FO Ori B filter data phased onto the 0.09 day period. 31

Figure 17. FO Ori B filter data phased onto the 0.097 day period after 0.09 day period removed. QY Aql QY Aql was observed on three separate nights using the 0.4-m telescope at Ball State. Analysis of the data acquired on the nights of 1-July-010, 15-July-010, and 9-Sep-010 show a period of 0.1119 + 0.004 day. A secondary period of 0.1030 + 0.000 day is found after removing the dominant period however the difference between the two frequencies is close to 1 c/d, implying this may be an artifact of 4-hour aliasing. Nonetheless, removal of this period from the period search reveals a secondary period at 0.1196 + 0.008 days. Figure 18 show the light curve in the V-Filter on the night of 1-July-010 showing amplitude of about 10 millimag. The photometric errors associated with that night were two millimag or less. Power-spectra for the 0.1119 and 0.1196 day periods are shown in figures 19 and 0. The data folded onto the 0.1119 day period is shown in figure 1. 3

-0.05-0.04-0.03 Δ Mag -0.0-0.01 55,378.7 55,378.76 55,378.80 55,378.84 55,378.88 55,378.9 0 JD +400000 Figure 18. Light Curve for QY Aql in the V filter on the night of 1-July-010 (UT). Figure 19. Power-spectrum of QY Aql data in the V filter with dominant peak at 0.1119 day. 33

Figure 0. Power-spectrum of QY Aql with peak at 0.1196 day after removal of 0.1119 day peak. Figure 1. QY Aql V filter data phased onto the 0.1119 day period. 34

SY Cen Of all the observed systems with periodicity detected, SY Cen is the most uncertain as to whether pulsation is truly present or not. Only one night of data has been acquired to date, and thus the periodicity has not been confirmed in multiple image sets, and the uncertainty in the measurement is quite large due to having only gathered one complete cycle. It is also the only target on the list for which a positive result for pulsation was detected and less than 00 images were acquired. It is strongly encouraged that further observations are obtained to confirm whether or not pulsations are indeed present or not. Nonetheless, PERANSO detected periodicity from the analysis of the 16 V-filter images acquired on 31-May-010 by the 0.6-m telescope. The light curve indicates amplitude of about 5 millimag. Photometric errors were less than 3 millimag throughout the night. The period was determined to be 0.0936 + 0.0119 day. Figures and 3 show the light curve and powerspectrum, respectively. No secondary peaks are found when removing the 0.0936 day period from the period search. This should not be surprising as observations were only taken one night. Also as a consequence, no 4 hour aliases are found in the power-spectra. 35

1.85 1.8 1.75 Δ Mag 1.7 1.65 1.6 1.55 55347.47 55347.5 55347.57 55347.6 JD +400000 Figure. SY Cen on the night of 31-May-010 (UT) in the V filter. Figure 3. Power-spectrum of SY Cen in the V filter on the night of 31-May-010. 36

WY Leo As mentioned above, WY Leo has already been published in the literature (Dvorak, 009) with a reported period of 0.0656 day. The author reported observations totaling 7 nights. The amplitude was determined to be 11 + 1 millimag. The observations for the earlier study were taken on a 0.5-m Meade Schmidt-Cassegrain with an SBIG ST-9XE CCD in the B and V filters. These observations are in contrast to those performed for the purposes of this thesis. As indicated in Table 1, observations on WY Leo were performed on the night of 9-March- 010. A total of 115 images were obtained on the 0.9 meter telescope in the V filter. The light curve showed definite variability, however the level of variability changed throughout the data set and thus definitive amplitude could not be determined. Photometric errors ranged between 1.5 and 1.8 millimag. The light curve is shown in Figure 4. The determined period for the observations obtained that night was 0.0457 +0.008 day (the power-spectrum is shown in Figure 5). The asterisk in Table next to the value of pulsation period indicates this differs from that obtained by Dvorak. Thus it is possible some mechanism has been affecting the system causing a period change. If this is the case, it would be a very significant period change in a short amount of time and would be worth further study. On the other hand, it is cautioned that for this thesis only one night of observations have been gathered, compared to the 7 nights by Dvorak. It is possible the period has not changed from the 0.0656 day obtained earlier, but not enough cycles were obtained in one night of observing to detect the dominant period. Thus, while further study of the object is encouraged, the caveat that one night of datataking should not be regarded as proof of change in the periodicity of the system. Data were 37

taken on the night of 15-Jan010 also from the 0.9-m telescope. However the quality of the data was not good enough to be included in this study. Because of the insufficient amount of data to show a change in periodicity, for the purposes of further analysis, the pulsation period of 0.0656 day obtained by Dvorak will be adopted. -0.1-0.1-0.08 mag diff -0.06 5584.6 5584.67 5584.7 5584.77 5584.8 5584.87-0.04 time Figure 4. Light curve for WY Leo on night of 9-March-010 (UT) in the V filter. 38

Figure 5. Power-spectrum for WY Leo in the V filter showing the 0.0457 day period. The residual peak is at 0.0831 day. X Tri Also as stated above, X Tri has also been previously observed. However detection of pulsation in the earlier study was negative (Liakos, 009). In their study, a 0.4-m Cassegrain telescope was used with an ST-8XMEI CCD camera. X Tri was observed on 13 separate nights in the B, V, and R filters. The observations covered a time span of 5.3 hours. X Tri was observed for this thesis on the nights of 6-Oct-010 and 7-Oct-010 using the 0.4-m telescope on campus. Over the two nights, 313 images in the V filter were obtained. These observations resulted in the lowest photometric errors obtained in this study, each measurement on 6-Oct-010 being less than one millimag and 7-Oct-010 were 1.3 millimag or less. The light curve for 6-Oct-010 is shown in Figure 6 (the data points do have error bars attached). The determined period of pulsation is 0.00 + 0.0001 day with amplitude of about 39

10 millimag. The power-spectrum is shown in Figure 7. Removal of the 0.00 day period did not reveal any significant residuals in the power spectrum. It is not believed that X Tri is multiperiodic, at least in the δ Scuti regime of periods. Figure 8 shows the data folded onto the 0.00 day period. 1.05 1.04 1.03 1.0 1.01 Δ Mag 1 0.99 0.98 0.97 0.96 0.95 55,475.74 55,475.78 55,475.8 55,475.86 JD +400000 Figure 6. Light curve of X Tri in the V filter for the night of 6-Oct-010 (UT). 40

Figure 7. Power-spectrum of X Tri in the V filter showing a peak at 0.00 day. Figure 8. Data for X Tri folded onto a single phase for the 0.00 day period. 41